diagnostic.visualizations.plot_metric_ranking#

plot_metric_ranking(df_metric, ax=None, plot_colorbar=True, hex_color1=None, hex_color2=None, **kwargs)[source]#

Plots a heatmap of the ranking of metrics for different model members.

This function takes a DataFrame of metrics, calculates the rankings of these metrics for each model member, and creates a heatmap representing the ranks. The plot can optionally include a colorbar to represent the ranking levels. If no axis is provided, a new figure and axis are created for the plot.

Parameters:#

df_metricpd.DataFrame

A DataFrame containing the calculated metrics for different model members. Each column represents a model member, and each row represents a metric.

axmatplotlib.axes.Axes, optional

A pre-existing axis to plot the heatmap. If None (default), a new figure and axis are created.

plot_colorbarbool, optional

If True (default), a colorbar is added to the plot to represent the rank levels. If False, the heatmap is plotted without a colorbar.

hex_color1str

The starting color of the colormap in hex format (e.g., ‘#FF0000’ for red).

hex_color2str

The ending color of the colormap in hex format (e.g., ‘#0000FF’ for blue).

Returns:#

axmatplotlib.axes.Axes

The axis object containing the heatmap plot.

Notes:#

  • The color map has the ‘summer’ palette as default and is resampled to the number of model members.

  • A customized color map can be included or determined as an interpolation between two colorcodes (hex codes)

  • Rankings are normalized based on the number of model members.

  • The function supports colorbar ticks to represent custom rank labels, which are added only if plot_colorbar=True.